mlampros / fasttextr Goto Github PK

Hi -
I'm following this tutorial ( https://cran.r-project.org/web/packages/textTinyR/vignettes/word_vectors_doc2vec.html
)

ERROR:

Error in skipgram_cbow(input_path = PATH_INPUT, output_path = PATH_OUT,  : 
  could not find function "skipgram_cbow"

Running into an error when running the following code:

`vecs` = fastTextR::skipgram_cbow(input_path = PATH_INPUT, output_path = PATH_OUT, 
                                 method = "skipgram", lr = 0.075, lrUpdateRate = 100, 
                                 dim = 300, ws = 5, epoch = 5, minCount = 1, neg = 5, 
                                 wordNgrams = 2, loss = "ns", bucket = 2e+06,
                                minn = 0, maxn = 0, thread = 6, t = 1e-04, verbose = `2)`

Thank you for your help!

@mlampros
thx for yr great job to bring fasttext into R
some R users like me, not so familar with python,
so can us show us some example to use fasttextR ?

Hi. I am trying to use skipgram_cbow command. And I am using Windows.
So, I installed fastTextR using devtook, but "fastTextR::skipgram_cbow(input_path = ...)" didn't work.

`> devtools::install_github('mlampros/fastTextR')

vecs = fastTextR::skipgram_cbow(input_path = PATH_INPUT, output_path = PATH_OUT,

•                             method = "skipgram", lr = 0.075, lrUpdateRate = 100, 
  

•                             dim = 300, ws = 5, epoch = 5, minCount = 1, neg = 5, 
  

•                             wordNgrams = 2, loss = "ns", bucket = 2e+06,
  

•                             minn = 0, maxn = 0, thread = 6, t = 1e-04, verbose = 2)
  

Error: 'skipgram_cbow' is not an exported object from 'namespace:fastTextR'`

So, I tried the following command, but got an error message when executing the fourth line:

`setwd('/your_folder/fastTextR/')
Rcpp::compileAttributes(verbose = TRUE)
setwd('/your_folder/')
system("R CMD build fastTextR")

• checking for file 'fastTextR/DESCRIPTION' ... OK
• preparing 'fastTextR':
• checking DESCRIPTION meta-information ... OK
• cleaning src
• checking for LF line-endings in source and make files and shell scripts
• checking for empty or unneeded directories
• building 'fastTextR_1.0.1.tar.gz'
  Warning in gzfile(tarfile, "wb", compression = compression_level) :
  cannot open compressed file 'C:/Program Files/R/R-3.5.1/library/fastTextR_1.0.1.tar.gz', probable reason 'Permission denied'
  Error in gzfile(tarfile, "wb", compression = compression_level) :
  cannot open the connection
  Execution halted

system("R CMD INSTALL fastTextR_1.0.0.tar.gz")`

"Permission denied" may mean that the command failed to make changes to the fastTextR folder since it is read-only. However, I tried to unlock the read-only option of the folder, it never gets unlocked. Or maybe I am pointing to a wrong issue. How do I get around with this error message?
Is there any other way to install fastTextR in Windows?

Thank you!

Error in .Call("fastTextR_convert_args_to_pointers", PACKAGE = "fastTextR", :
"fastTextR_convert_args_to_pointers" not available for .Call() for package "fastTextR"

I compiled the package from the repo. But for some reason I get this error when I try to use the package. Any idea how to solve this?

Hi @mlampros,

Is there a way to use FastTextR to predict the top N number of similar documents? I've gone through the algorithm and can compile the doc2sum with k means (per your example), but having trouble being able to get top 5 most related to a single document.

Thanks for your help.

Hi. I am following the example presented below:
https://cran.r-project.org/web/packages/textTinyR/vignettes/word_vectors_doc2vec.html

At the stage of using fasttextR command as below:
vecs = fastTextR::skipgram_cbow(input_path = PATH_INPUT, output_path = PATH_OUT,
method = "skipgram", lr = 0.075, lrUpdateRate = 100,
dim = 400, ws = 5, epoch = 10, minCount = 1, neg = 5,
wordNgrams = 2, loss = "ns", bucket = 2e+06,
minn = 0, maxn = 0, thread = 6, t = 1e-04, verbose = 2)

While this code above is being executed, R presents loss values for each epoch.
However, after the execution is over, those verbose containing loss values are gone.
Is there any way for me to access the values of loss for each epoch?
I am trying to capture these loss values in order to decide the optimal size of the epoch by figuring out starting from which epoch the loss flattens out.

Additionally, is it possible to compute the loss value after training a model with a specific epoch size?

Looking forward to hearing from you soon. Thank you!

Hi,

I want to install the Fasttext package from computer of my company. Due to the company's firewall, I have to manually install this way:

setwd('C:/your_folder/fastTextR/')
Rcpp::compileAttributes(verbose = TRUE)
setwd('C:/your_folder/')
system("R CMD build fastTextR")
system("R CMD INSTALL fastTextR_1.0.2.tar.gz")

Even through I tried to install Rtools but it still does not work:

============================================================

Configuration CCC :

Proxy :
    HTTP  = pxy-http-srv:1234
    HTTPS = pxy-http-srv:1234

Serveur de paquets (le seul accessible via le proxy) :
    https://mirror.ibcp.fr/pub/CRAN/

Librairie locale de paquets : 
    C:\Developpement\R_Packages\

Fichiers de config pris en compte :
    C:/PROGRA~1/R/R-35~1.0\Renviron.site
    C:/PROGRA~1/R/R-35~1.0\Rprofile.site

Error in library("colorout") :
no package named 'colorout' is found
Stopped execution
In R CMD INSTALL

Could someone please tell me how can I solve this problem?

Thanks a lot for your information!